Sweep: AI Coding Built Specifically for JetBrains

Sweep is an AI coding assistant for JetBrains IDEs that provides context-aware autocomplete, AI code editing, and project-wide code suggestions.

Description

Sweep is an AI coding assistant built specifically for JetBrains IDEs, including IntelliJ IDEA, PyCharm, WebStorm, GoLand, PhpStorm, Rider, and CLion. It combines fast next-edit autocomplete, powered by a custom 1.5B-parameter model, with an AI agent that can edit multiple files across a codebase to handle refactoring, bug fixes, and code review. Sweep indexes the Program Structure Interface (PSI) of a project, so it understands classes, methods, and variables directly rather than relying on external embeddings or slow cloud indexing. Autocomplete is unlimited on paid plans, while agent tasks like multi-file edits and code review draw from a monthly pool of API credits that rolls over for 30 days. A Privacy Mode setting ensures code is used only to generate suggestions and never stored or used for model training. It suits developers working inside JetBrains IDEs who want an AI assistant with deep codebase understanding rather than a general-purpose plugin ported from another editor.

Key Features

  • Next-edit autocomplete — predicts intended edits and suggests changes in milliseconds, unlimited on paid plans.
  • Multi-file AI agent — edits across multiple files in a codebase for refactoring and bug fixes.
  • PSI-based indexing — understands a project’s classes, methods, and variables directly rather than through external embeddings.
  • AI code review — automated review and commit message generation.
  • Privacy Mode — code is used only for suggestions and never stored or used for training.
  • Credit rollover — unused API credits roll over for 30 days rather than expiring immediately.
  • Team plan — adds SSO, usage visibility, and organization-wide Privacy Mode enforcement.

How It Works

Sweep installs as a plugin from the JetBrains Marketplace into any JetBrains IDE. Once installed, next-edit autocomplete runs continuously as a developer types, predicting the next likely edit based on the project’s indexed structure. For larger tasks, the AI agent can be directed to make multi-file changes, refactor code, or review a pull request, drawing from a monthly allocation of API credits included with a paid plan; purchasing a plan grants that dollar amount in included usage, and any additional usage beyond it is billed at the underlying LLM’s cost plus a small markup. Unused credits carry over for 30 days rather than resetting immediately. Team plans add centralized billing, usage visibility across the organization, and the ability to enforce Privacy Mode for every member rather than per individual.

Technical Architecture & Overview

  • Core Engine: Custom 1.5B-parameter model for autocomplete, paired with larger LLMs for agent tasks like multi-file edits and code review.
  • Deployment: JetBrains plugin, installed from the JetBrains Marketplace; runs inside the IDE rather than as a separate application.
  • API Surface: No public developer API documented; usage is metered through in-IDE API credits rather than an external API key.
  • Known Limits: Credits beyond a plan’s included allowance are billed at the underlying LLM’s cost plus a markup; when a Team seat runs out, a new seat is purchased automatically for the next user who joins.

Pros & Cons

ProsCons
Built specifically for JetBrains IDEs, rather than a general plugin ported from VS CodeOnly available for JetBrains IDEs; no support for VS Code or other editors
PSI-based indexing gives direct understanding of a project’s structure without slow external embeddingsAgent tasks beyond autocomplete draw from a metered credit pool rather than being fully unlimited
Unused API credits roll over for 30 days instead of resetting immediatelyOverage usage beyond included credits is billed at LLM cost plus a markup
Privacy Mode guarantees code is never stored or used for model training when enabledTeam seats are purchased automatically when new members join, which can surprise budgets without monitoring
Discounts available for students, educators, and open-source maintainers on requestNo published free tier for individual (non-open-source) developers

Pricing

PlanPrice
Pro$20/month ($20 in included API credits, unlimited autocomplete)
Ultra$60/month (higher API credit allocation, priority support)

https://sweep.dev/pricing

Platform Availability

JetBrains IDEs (IntelliJ IDEA, PyCharm, WebStorm, GoLand, PhpStorm, Rider, CLion, RubyMine)

Best For

JetBrains IDE users | Backend and full-stack developers | Development teams

Frequently Asked Questions

Which JetBrains IDEs does Sweep support?

Sweep works with all JetBrains IDEs, including IntelliJ IDEA, PyCharm, WebStorm, GoLand, PhpStorm, Rider, CLion, and RubyMine, installed directly from the JetBrains Marketplace, per Sweep’s official site.

What’s the difference between Pro and Ultra?

Pro costs $20/month with $20 in included API credits and is aimed at power users, while Ultra costs $60/month with a higher credit allocation for developers who want maximum agent usage, per Sweep’s official pricing page.

Does Sweep train on my code?

Not when Privacy Mode is enabled. With Privacy Mode on, code is used only to generate AI suggestions and is never stored or used for model training, according to Sweep’s official site.

Do unused Sweep credits roll over?

Yes. Unused API credits roll over for 30 days rather than resetting at the start of each billing period, per Sweep’s official pricing documentation.

Are there discounts for students or open-source maintainers?

Yes. Sweep offers special pricing for students, educators, and open-source maintainers who contact the team directly with verification of their status.

Visit Official Website

- Advertisement -

Latest